Integrated STEM & Career Education for Middle Schools

Ecological Applications, 24(5), 2014, pp. 1101–1114� 2014 by the Ecological Society of America

Contrasting evolutionary demography induced by fishing:the role of adaptive phenotypic plasticity

MANUEL HIDALGO,1,6 ESBEN M. OLSEN,1,2 JAN OHLBERGER,1 FRAN SABORIDO-REY,3 HILARIO MURUA,4

CARMEN PINEIRO,5 AND NILS C. STENSETH1,3

1Centre for Ecological and Evolutionary Synthesis (CEES), Department of Biosciences, University of Oslo, P.O. Box 1066 Blindern,0316 Oslo, Norway

2Institute of Marine Research, Flødevigen Marine Research Station, 4817 His, Norway3Institute of Marine Research (IIM-CSIC), Eduardo Cabello 6, 36208 Vigo, Spain

4AZTI Tecnalia, Herrerra Kaia, Portualde z/g, Pasaia (Guipuzkoa), Spain5Instituto Espanol de Oceanografıa, Centre Oceanografico de Vigo, Cabo Estay, Canido, Apartado 1552, 36200 Vigo, Spain

Abstract. Mounting evidence now shows that fishing activity modifies both heritable life-history traits and ecological processes in harvested populations. However, ecological andevolutionary changes are intimately linked and can occur on the same time scale, and fewstudies have investigated their combined effect on fish population dynamics. Here, we contrasttwo population subunits of a harvested fish species in the Northeast Atlantic, the Europeanhake (Merluccius merluccius), in the light of the emerging field of evolutionary demography,which considers the interacting processes between ecology and evolution. The two subunitsexperienced similar age/size truncation due to size-selective fishing, but displayed differences inkey ecological processes (recruitment success) and phenotypic characteristics (maturationschedule). We investigate how temporal variation in maturation and recruitment successinteractively shape the population dynamics of the two subunits.

We document that the two subunits of European hake displayed different responses tofishing in maturation schedules, possibly because of the different level of adaptive phenotypicplasticity. Our results also suggest that high phenotypic plasticity can dampen the effects offisheries-induced demographic truncation on population dynamics, whereas a populationsubunit characterized by low phenotypic plasticity may suffer from additive effects ofecological and life-history responses. Similar fishing pressure may thus trigger contrastinginteractions between life history variation and ecological processes within the samepopulation. The presented findings improve our understanding of how fishing impacts eco-evolutionary dynamics, which is a keystone for a more comprehensive management ofharvested species.

Key words: contemporary evolution; demographic erosion; European hake; evolutionary demography;fisheries conservation; fishing-induced effects; Merluccius merluccius; phenotypic plasticity.

INTRODUCTION

Understanding how fishing modifies the natural

mechanisms that regulate fish populations’ productivity

is of serious concerns for fisheries ecologists (Rose et al.

2001, Lorenzen and Enberg 2002). This understanding is

a prerequisite for designing stock-specific conservation

strategies and for better predicting the future function-

ing of marine ecosystems (Fisher et al. 2010). Particu-

larly, natural mechanisms of regulation can be modified

by the disproportional removal of older and larger

individuals due to fishing activity (Law 2000). The view

embodied in fisheries science is that variation of adult

density is the main factor regulating the juvenile

survival, while changes in phenotypic characteristics of

adults have been investigated as independent processes

(Lorenzen and Enberg 2002). However, phenotypic

variation can have an essential role in the population

regulation of harvested species (Goodwin et al. 2006,

Marshall et al. 2010).

The effects of size-selective fishing on both population

dynamics and life history have been documented

worldwide (for a review, see Jørgensen et al. [2007],

Hsieh et al. [2010]). For instance, the erosion of the age

structure makes populations more dependent on youn-

ger age classes (i.e., demographic erosion; Ottersen

2008), which tightens the link between population and

environmental variability (Anderson et al. 2008). Size-

selective exploitation also affects population dynamics

through diminishing maternal influences, which reduces

the positive effect of size-related reproductive traits

(fecundity, quality of eggs, or bet-hedging strategies) on

the offspring survival (Venturelli et al. 2010). Further-

more, there is accumulating evidence that selective

Manuscript received 15 October 2013; revised 19 October2013; accepted 10 December 2013; final version received 6January 2014. Corresponding Editor: O. P. Jensen.

6 Present address: Instituto Espanol de Oceanografıa,Centre Oceanografic de les Balears, Moll de Ponent s/n,07015 Palma, Spain. E-mail: [email protected]

1101

fishing also causes evolutionary changes in fitness-

related traits like growth and maturation (Olsen et al.

2004, Swain et al. 2007). However, few studies have

investigated how fishing-induced changes in life history

and demography interact to shape population dynamics.

This remains a serious challenge for fisheries ecology.

Evolutionary demography, or eco-evolutionary dy-

namics, aims to integrate ecological responses in

evolutionary studies, as well as evolutionary responses

in ecological studies, to comprehend the interplay

between these processes (Pelletier et al. 2007, Ezard et

al. 2009). Phenotypic plasticity (i.e., the ability of a

genotype to produce different phenotypes across an

environmental gradient) can modify this interplay due to

its potential to reduce evolutionary change (Price et al.

2003, Reed et al. 2010, Ellner et al. 2011). This is of

paramount importance in exploited marine populations

because phenotypic plasticity may constrain the evolu-

tionary impact of human-induced selection (Law 2000,

Marshall and McAdam 2007). In addition, individuals

inhabiting heterogeneous environments are more prone

to display plastic changes in their life history traits (Reed

et al. 2010). This is the case in spatially structured

marine populations, in which subunits (i.e., demes) can

be affected differently by environmental heterogeneity

(Kareiva 1990). In this case, evolutionary demography

may differ for subpopulations colonizing environments

of different complexity, provided that local selection

pressures outweigh the homogenizing effect of gene flow.

To investigate the influence of fishing in shaping

evolutionary demography, a first step is to study

populations or sub-populations of the same species.

We selected a long-term exploited species in the

Eastern Atlantic (European hake, Merluccius merluc-

cius; see Plate 1) that exhibits a complex spatial structure

with two main demes consistent with two current

management units (i.e., stocks). These demes are not

genetically different (Pita et al. 2011) but display

spatially independent and temporally stable spawning

aggregations (Fig. 1a), and distinct recruitment dynam-

ics (Hidalgo et al. 2012; see also Material and Methods).

Thus, they were investigated as independent demes

within a population. The two demes were historically

affected by the same size-selective pattern of fishing

mainly targeting the adult stock (ICES 2009, Murua

2010), displaying similar temporal trends in population

biomass (Fig. 1b) and a progressive demographic

erosion of age structure (Fig. 1c). By contrast, temporal

trends in length at maturity differed considerably

between the two demes (Fig. 1d; Domınguez-Petit et

FIG. 1. (a) Area covered by the north (NA, black square) and south (SA, gray square) management units of hake investigatedas independent demes (Pita et al. 2011). Light and dark gray areas of the ocean indicate the main nursery and spawning areas,respectively; GS and CS refer to Galician shelf and Celtic Sea (see Materials and Methods for study system description). Graphsshow temporal variation of (b) deme biomass, (c) mean age of spawners (spawning stock biomass, SSB) and (d) lengths at 50%probability of being mature (error bars represent SE). Black and gray lines refer to the north and the south demes, respectively.

MANUEL HIDALGO ET AL.1102 Ecological ApplicationsVol. 24, No. 5

al. 2008). These characteristics make hake a suitable case

study to investigate interactions between ecological

processes (i.e., recruitment) and life history dynamics

(i.e., maturation) on population dynamics, and whether

fishing alters this interaction.

The underlying hypothesis is that the level of

phenotypic plasticity in length at maturation shapes

the interaction between phenotypic changes and ecolog-

ical processes. To understand how fishing modifies this

interaction, we synthesize and combine results from

three modeling techniques within an evolutionary

demography framework (Fig. 2). First we investigate

temporal changes in the phenotype, estimating matura-

tion reaction norms for each deme to elucidate the

relative contribution of adaptive changes vs. phenotypic

plasticity in the maturation schedules. Second, we

examine the temporal changes in the recruitment process

by analyzing variation in recruitment success and how

the density-dependent regulation in stock-recruitment

models changes over time. Finally, we used age-

structured matrix models to investigate the population

dynamics consequences of the interaction between

changes in the recruitment success and those in

maturation schedules. Matrix models are well suited to

investigate this interaction since one of the vital rates of

the matrix structure is the reproductive rate, which

explicitly results from the combination of maturation

rates and recruitment success (see Materials and

Methods and Fig. 2).

MATERIALS AND METHODS

Two contiguous contrasting systems

The two contiguous demes of the hake population in

the Northeast Atlantic inhabit contrasting environments

(e.g., hydrographical characteristics) that results in

different degrees of environmental and population

heterogeneity (Fig. 1a). The north deme displays a large

spawning area along the shelf break of the broadcontinental shelf of the Bay of Biscay, while the Celtic

Sea being the main nursery area. The large-scale

hydrography in the Bay of Biscay controls the zoo-plankton dynamics and the recruitment success of hake

in the whole north deme (Goikoetxea and Irigoien

2013). By contrast, the south deme is distributed alongthe Atlantic Iberian coast and the Cantabric Sea, which

are both characterized by a narrow continental shelf.

This makes spatial distributions more patchy comparedto the north deme. The spawning areas are mainly

located in the Galician waters (Domınguez-Petit et al.

2008) while the nursery areas are often found aroundorographic structures such as big capes where primary

and secondary production locally increases (Sanchez

and Gil 2000). Thus, small hydrographical processes

such us Iberian upwelling (Domınguez-Petit et al. 2008)or the strength of shelf-edge currents (Sanchez and Gil

2000) are the main environmental drivers influencing the

recruitment dynamics at the south deme.The window of data analyzed in the present study

includes two periods of contrasting demography (mean

age, Fig. 1c) of the spawning stock as well as differentlengths at 50% probability of being mature (Fig. 1d).

Both the north and the south deme experienced a similar

progressive truncation of age structure that is maxi-

mized around the early nineties in both stocks (Hidalgoet al. 2012). The strategy of the present study is,

therefore, to compare two periods that are contrasted

in terms of age structure and lengths at 50% probabilityof being mature in both population subunits. For

comparative analytical purposes (i.e., a balanced num-

ber of years between periods) and synthesis of the resultsbetween population subunits, two periods were selected:

before 1995 and from 1996 onward. Note that the lack

of maturity-at-length (see Life history and demographic

FIG. 2. Evolutionary demography framework applied combining information from three modeling techniques (probabilisticmaturations reaction norms, stock–recruitment relationships, and age-structured matrix models) in a combined and integrativemanner. Key information to be discussed and analyzed for each technique within an evolutionary demography framework isspecified. Note that temporal variation of ecological and evolutionary properties are investigated in the present study applyingtime-varying analytical approaches, with the exception of the reaction norms due to data restrictions (see specific methodology inMaturation reaction norms).

July 2014 1103EVOLUTIONARY DEMOGRAPHY IN FISH STOCKS

information section hereafter) in a critical period

between 1991 and 1995 influences the selection of these

two periods. Given the contrasting age structure

between periods for the two subunits, we refer to the

time period before 1995 as the ‘‘pre-truncated period’’

and to the period from 1996 onward as the ‘‘truncated

period.’’

Life history and demographic information

Maturity was macroscopically assessed using gonads

sampled monthly in the main spawning areas from 1983

to 2008 for the south deme and for 1987–1990 and 1996–

2008 for the north deme (details in Domınguez-Petit et

al. 2008). Age-structured abundance and spawning stock

biomass were extracted for the period 1978–2008 (north

deme) and 1982–2008 (south deme) from the hake

assessment of the International Council for the Explo-

ration of the Sea (ICES 2009). These ICES groups used

virtual population analyses with extended survivor

analysis (VPA-XSA) to estimate abundance-at-age

based on commercial catch data tuned to survey data.

In these assessment procedures, no stock–recruitment

relationship is assumed. Note that annual estimates of

maturity-at-age for the north deme are not available for

two periods: 1978–1986 and 1991–1995. For population

dynamics purposes of this subunit, the maturity-at-age

matrix of 1986 was extended backward, while a weighted

estimation was applied to interpolate between 1991 and

1995. To do that, we considered a progressive contribu-

tion of the information of the two years on the edge

(1990 and 1996), applying a weighting vector (0.16, 0.33,

0.5, 0.66, 0.83) to calculate the maturity-at-age of the

five unavailable years.

Maturation reaction norms

We applied the probabilistic maturation reaction

norms approach to compare the reaction norm mid-

points (i.e., combinations of age and size for which the

estimated probability of maturing is 50%) between the

two contrasting periods. This methodology can help to

disentangle phenotypically plastic and evolutionary

changes in maturation schedules (Barot et al. 2004,

Olsen et al. 2004). Probabilistic reaction norms for age

and size at maturation are defined by the age- and size-

specific probabilities at which individuals within a

cohort mature during a given time interval (Heino et

al. 2002), which overcomes the confounding effects of

growth and mortality (often environment dependent).

The probability of maturing at a given age and size

(m(a,s)) was calculated from the probability of being

mature at that age and size (i.e., the maturity ogive,

o(a,s)) and the probability of being mature at the

previous age and size (o(a � 1, s � Ds(a))), which was

inferred on the mean annual growth increment (Ds):m(a,s) ¼ [o(a,s) – o(a � 1, s – Ds(a))]/[1 – o(a � 1, s –

Ds(a))]. To compare periods with contrasting demo-

graphic structure (Fig. 1c) and different lengths at 50%probability of being mature (Fig. 1d), information of the

pre-truncated and of the truncated period was pooled

for each of the two demes. For each deme and period we

calculated the probabilistic reaction norm midpoints for

age-2 and age-3, corresponding to the size classes at

which most of the fish mature. We expect that large

differences between reaction norms of each period could

suggest a potentially higher contribution of evolutionary

changes to the maturation schedules compared to

phenotypically plastic variation.

Age estimations for hake have an inherent uncertainty

due to the difficulty of interpretation of the annual rings

of the otoliths (ICES 2009, Murua 2010). This would

bias our reaction norm estimates due to the sensitivity of

the probabilistic reaction norms to changes in somatic

growth (Dieckmann and Heino 2007), and thus prevent

of the calculation of probabilistic maturation reaction

norms by cohort. To circumvent this limitation and to

be able to evaluate a potential effect of changes in

somatic growth on the reaction norms, we adopted the

strategy of testing the sensitivity of the reaction norms

estimated to simulated variation in hake aging using a

range of somatic growth rate scenarios (Ohlberger et al.

2011). Particularly, we used simulations to identify the

most likely somatic growth scenarios (higher, equal, or

lower) in the pre-truncated period compared to the more

recent period, which was used as known reference (see

simulations in Appendix A). These growth scenarios

illustrate the different contributions of plastic and

adaptive mechanisms at each population subunit by

allowing a comparison of the maturation reaction norm

midpoints between periods. Once the most likely

somatic growth scenario was identified for each popu-

lation subunit, we averaged information for age-2 and

age-3 to present the maturation reaction norm mid-

points for each population subunit and for each of the

selected periods.

Non-stationarity of population regulation

Growing evidence suggests that stock–recruitment

relationships (SRs) of fish populations are not tempo-

rally static (Enberg et al. 2010, McClatchie et al. 2010,

Olsen et al. 2011). Ecological characteristics such as

environmental influences, density dependence or steep-

ness can change with time at different time scales. We

here investigate potential temporal changes in the

density-dependent regulation of the recruitment success.

We particularly examine changes in the density-depen-

dent parameter of SRs that has been observed to

temporally correlate with changes in life history

characteristics (Goodwin et al. 2006). Several studies

highlight the utility of stock-recruitment relationships to

investigate density-dependent processes (e.g., survival;

Minto et al. 2008) and their relationships with life

history (Goodwin et al. 2006).

As preliminary analyses we compared the fit over the

whole time series of the two most common SRs

assuming density-dependent regulation in order to

identify which model best explains SR: the Ricker


model (overcompensation regulation) and the Beverton-

Holt model (saturation regulation; see Appendix B). The

Ricker model displayed a better fit for the two demes

over the whole period (Appendix B). We then applied a

12-year moving window fitting the linearized transfor-

mation of the model: log(Rt/SSBt�1) ¼ a0 – b(SSBt�1) þet, where Rt is the recruitment (age-0) at year t, SSBt�1the spawning abundance at year t � 1, a0 and b are the

estimated parameters, and et the error term at year t. To

analyze the temporal variation of the density-dependent

parameter (b), we applied the nls function provided in R

2.12.1 (R Development Core Team 2010), which uses the

Gauss-Newton algorithm to minimize the nonlinear

least-squares estimates of the model parameters. Note

that the size of the window was arbitrarily fixed to 12

years, which represents a compromise between the

limited length of the time series and the generation time

of the species, 5.5 years (T. Rouyer, personal communi-

cation). We expect that a significant temporal drift in b

would reflect a change in the density-dependent regula-

tion of recruitment within population subunit.

Population dynamics consequences

We investigated the population dynamics implications

of the interaction between maturation schedules and

recruitment success by calculating the temporal varia-

tion in population growth (r), as well as the relative

contribution (i.e., elasticity) of the reproductive rates to

changes in r. We used an age-structured matrix model

(Caswell 2001), which summarizes age-specific vital

rates: survival (S ) and the reproductive rate (RR). The

fundamental relationship of age-structured matrix mod-

els is given by Ntþ1 ¼ AtNt, where Nt is a vector

representing the number of individuals for each age class

at time t, and At is a transition (Leslie) matrix that

summarizes the dynamics of the populations between

time t and time t þ 1 (Leslie 1945, Caswell 2001). At

summarizes the vital rates and the logarithm of its

dominant eigenvalue (k) provides the population growth

rate, r. For a given year, the transition matrix (At) is

defined as follows:

At ¼

RR0;t RR1;t RR2;t . . . RRimax;t

S1�0;t 0 0 . . . 0

0 S2�1;t 0 . . . 0

. . . . . . . . . . . . . . .0 0 0 Simax�ðimax�1Þ;t 0

0BBBB@

1CCCCA

ð1Þ

with Si�(i�1),t the survival between age-class i at year t

and age-class i – 1 at year t – 1, and with RRi,t the

reproductive rate of the year-class i at year t defined as

follows:

RRi;t ¼RtMati;t�1

Xi¼imax

i¼1

Mati;t�1Ni;t�1

¼ RSt 3 Mati;t�1 ð2Þ

where Mati,t�1 is the maturation rate at age a and at time

t� 1, Ni,t�1 the abundance at age-class i and at time t�1, and Rt the recruitment at time t. Therefore, RRi,t

represents the key component of the present study

because results from the interaction between Mat and

the recruitment success (RSt, recruitment/(total abun-

dance of spawners), Rt/Nsp,t�1), which is an annual value

independent of age. Though the two demes displayed a

similar trend in density (Fig. 1a), we had preliminary

evidence of a contrasting non-stationary pattern of

density dependence between demes. Therefore, density

dependence was not explicitly included in the matrix

model to facilitate the inter-deme comparison.

We also computed the elasticity (E) of population

growth rate to changes in reproductive rate following

methodology provided in Caswell (2001). E has been

shown to provide direct information on the effect of

size-selective fishing on the population performance by

modifying the relative contribution of the reproductive

rate to the annual variation of the population growth

(Rouyer et al. 2011, Hidalgo et al. 2012). Since the

reproductive rate results from the interaction of Mat

and recruitment success, we specifically aim at disen-

tangling the influence of the changes in Mat on the

population dynamics from the ecological effects of the

recruitment success. To do that, we applied two

complementary approaches.

First, we compared the realized growth rate (r) with

the capability of a population to grow independent of

the recruitment success variability. To do that, recruit-

ment-independent growth rate (rRI) was calculated using

a ‘‘partial life cycles analyses’’ (i.e., demographic models

in which part of the age-specific estimates are lacking,

e.g., Oli and Zinner [2001]). In these analyses, the

reproductive rate was replaced by Mat in the Leslie

matrix with recruitment success remaining at a constant

value of 1 for the whole study period. The parameter rRI

can easily track the influence of changes in Mat (Hidalgo

et al. 2012), and contributes to identify the absolute

contribution of the recruitment success to r depicted by

the difference between r and rRI. We also computed

elasticity of rRI to Mat variation to be compared with

elasticity of r to reproductive rate variation. To

investigate whether an increasing or decreasing pattern

of elasticity led to an overall influence on the population

growth rate trajectories at short-term scale, we applied

the 12-year moving window previously described to

calculate Pearson correlations between each type of

growth rate and their respective elasticity to reproduc-

tive-related rates. The relationship between elasticity

values and population growth rate has been previously

used to demonstrate that population growth is not

uniformly sensitive to demographic rates across different

environmental conditions and demographic states

(Ezard et al. 2008). We ask, in the present study,

whether this causative relationship changes over time

according to a change in the demographic state of the

sub-populations. We expect a change in the sign of the

correlation between elasticity and r (or rRI) from the pre-


truncated period to the truncated period that would

indicate a switch in the control of the population

dynamics from a survival-dependent to a reproductive-

rate (or Mat, in the case of rRI)-dependent population

dynamics. The number of degrees of freedom was

adjusted to account for autocorrelation in all the

correlations performed (Pyper and Peterman 1998).

Second, we focus on the dynamics of the reproductive

rate by evaluating the relative importance of changes in

heritable phenotypic traits (i.e., Mat) and ecological

processes (i.e., recruitment success) affected by fishing in

influencing the population dynamics. To do that, we

applied Hairston et al.’s framework (Hairston et al.

2005), which aimed at comparing ‘‘ecological and

evolutionary dynamics.’’ The rationale behind this

framework is that temporal changes in some attributes

of the population dynamics are the consequence of

temporal changes in ecological and evolutionary pro-

cesses (Hairston et al. 2005, Ezard et al. 2009, Ellner et

al. 2011). The simplified mathematical expression in

discrete time to one ecological and one evolutionary

effect is

Xðt þ hÞ � XðtÞ ¼ ]X

]kDk þ ]X

]zDzþ . . . ð3Þ

where X is an attribute of the population dynamics, k is

an ecological variable, z is an evolutionary variable, h is

the interval between samples and ‘‘. . .’’ are higher-order

terms including interactions between the changes in the

driving variables. In our study, we focus on the two

population dynamics attributes (X ) that will contribute

to understand the evolutionary demography of harvest-

ed fish populations because they are shaped by the

interaction of ecological and evolutionary variables

possibly influenced by fishing. These attributes were

the reproductive rate and the elasticity of r to changes in

the reproductive rate (i.e., contribution of reproductive

rate to r).

Hairston et al. (2005) was the seminal work providing

a framework to investigate the consequences of tempo-

ral changes in ecological and evolutionary processes on

population dynamics attributes. However, Ellner et al.

(2011) have recently updated Hairston et al.’s approach

to disentangle ecological impacts of evolutionary change

vs. non-heritable trait change to avoid misestimating the

contribution of the ecological (k) and the evolutionary

variable (z). However, since heritability information

related to hake traits was not available and because our

aim was to compare the two demes, we do not attempt

to quantify the specific contributions and we adhere to

Hairston et al.’s approach. Thus, z is hereafter referred

to as a phenotypic variable in our study. In addition,

note that we do not aim at investigating the effect of the

environment in the present study, so we retain the global

ecological effect (k), which includes demographic and

environmental effects. We used age at 50% probability

of being mature as z, a heritable phenotypic trait that

summarizes the maturation schedules, and the recruit-

ment success as ecological variable (k). To quantify the

relative contributions of k and z on X, we apply ageneral linear model for each deme in which k and z are

treated as additive or interactive covariates (Hairston etal. 2005). The Akaike information criterion (AIC) was

used to select between the additive and the interactivemodel. We applied this modeling framework to thecomplete period, to the pre-truncated (before 1995) and

the truncated period (after 1996). We expect differenteffects of z and k on X for each deme at each period that

would indicate a contrasting evolutionary demographywithin the population (Ezard et al. 2009). For every

model, residuals were checked for homogeneity ofvariance, absence of temporal autocorrelation and

violation of normality assumptions.

RESULTS

Maturation reaction norms

For the north deme, the most likely growth scenariowas similar between periods (Appendix A), suggesting

low phenotypic plasticity in somatic growth. Based onthis growth pattern, maturation reaction norms had a

negative slope and thus older individuals are more likelyto mature than younger ones at a given size (Fig. 3a).

Further, maturation reaction norms differed betweenperiods, with midpoints for the truncated period

considerably shifted toward young age and small sizecompared to the pre-truncated period (Fig. 3a). This

suggests an evolutionary change in the maturityschedules of the north deme.

In contrast, the south deme displayed a highdependency on somatic growth with a lower growth

rate in the pre-truncated than in the more recenttruncated period (see Appendix A). This suggests a

higher phenotypic plasticity in somatic growth in thesouth compared to the north deme. This also explains

the uncertainty of the reaction norm estimate at laterages during the truncated period, because fish grew

faster and matured at age-2 thereby reducing thenumber of first maturing individuals at age-3 (Fig. 3b).The resulting averaged maturation reaction norms (Fig.

3b) did not vary significantly between periods indicatingno influence of evolutionary processes on maturity

schedules, which were instead likely shaped by plasticresponses in growth. In addition, the low variation

between periods for age-2 and the overlapping standarddeviation for age-3 suggest a maturation reaction norm

less variable in terms of size and more independent ofage compared to the north deme.

Non-stationary density dependence in stock-recruitmentrelationships

The Ricker model applied over the whole time series

indicated a similar general pattern for the two demeswith significant density dependence (b ¼ 4.55 3 10�6 6

5.86 3 10�7 [mean 6 SE], P � 0.001, Fig. 4a for the

north deme; b ¼ 3.59 3 10�5 6 9.55 3 10�6, P , 0.001,Fig. 4b for the south deme). However, clear differences


FIG. 3. Midpoints of the probabilistic maturation reaction norm for age-2 and age-3 estimated for the pre-truncated period(black lines) and truncated period (gray lines), and for (a) north deme and (b) south deme. Error bars indicate SD. Note that, forthe pre-truncated period (black lines), midpoints are calculated based on the most likely growth scenarios obtained fromsimulations (Appendix A): similar somatic growth between periods for the north deme while lower somatic growth in the pre-truncated than in the truncated period for the south deme.

FIG. 4. (a, b) Relationships between spawning stock biomass (SSB) and recruitment at age-0 (R) for the (a) north and (b) southdeme. Solid lines represent the predicted Ricker model for the whole time series, while the color of the circles represents the meanage of spawners (lighter is lower, darker is higher). Dashed-dotted and dotted lines represent the predicted Ricker model for thepre-truncated (before 1995) and post-truncated (after 1996) period respectively. (c, d) Density-dependent parameter (b; open blackcircles) and coefficient of variation of log recruitment success (R/SSB; gray lines) estimated applying a 12-year moving-window.Vertical lines indicate SD and the symbols indicate the significance of b (�P , 0.005 in c and * P , 0.05 in d). Horizontal lines (solidline, mean; broken lines, SD) indicate estimated b values for the whole period from Fig. 4a and 4b.


appeared when the model was fit for each period with

different demographic structure (Fig. 4a, b). Density

dependence calculated using a 12-year moving widow

displayed a strong and stable effect on the recruitment

dynamics of north deme until the early 1990s, consistent

with the b value of the complete time series. After that,

density dependence rapidly disappeared, presumably

due to the fishing-induced demographic erosion (Fig.

4c), though the low range of spawning abundance in the

truncated period of the north deme does not allow to

properly defining a stock–recruitment relationship.

Nonsignificant values of b were generally consistent

with low values of the coefficient of variation (calculated

with the same moving widow) of the recruitment success

(Fig. 4c). By contrast, density dependence does not seem

to influence the recruitment dynamics at the south deme

during the first part of the study period when the age

structure was older (Fig. 4d). Density dependence

increased (i.e., b became significant) as the demographic

truncation progressed toward the recent years concom-

itant with an increase of the coefficient of variation of

the recruitment success (Fig. 4d). This indicates that the

b parameter of the whole time series for the south deme

represents an average value of the transition from

absence of density-dependent regulation in the pre-

truncated period to that recovered in the recent period

(Fig. 4d).

Population growth rates and contribution of reproductive-

related vital rates

Realized growth rate (r) and the relative contribution

(i.e., elasticity) of the reproductive rate to changes in r

were estimated from transition matrices to investigate

the population dynamic implications of changes in the

maturation schedules and in the recruitment success.

Time series of r displayed, in general, negative values for

the two demes, which indicate the simultaneous tenden-

cy in these subpopulations to decrease until early 2000s

(Fig. 5a). By contrast, recruitment-independent growth

rate (rRI, capability of a population to grow independent

of the recruitment success variability) was always lower

compared to r for the south deme (Fig. 5b), while rRI

and r were quite similar for the north deme (Fig. 5a). In

addition, the variability of the absolute contribution of

the recruitment success (i.e., offspring survival, r � rRI)

increased through the last decades for the south while

that for north deme was cyclic and relatively stable

around zero (Fig. 5a). The parameter rRI was thus used

PLATE 1. European hake (Merluccius merluccius) at the Galician (northwestern Spain) waters. European hake is a long-termexploited demersal species at both the North East Atlantic and Mediterranean waters. Photo credit: Jorge Hernandez Ucera.


to disentangle the influence of the changes in maturity

rates from the ecological effects of the recruitment

success. These results suggest a high influence of the

annual offspring survival on the population perfor-

mance of the south deme compared to the north deme.

In both population subunits, rRI and r became more

elastic to (i.e., relatively more dependent on) the

respective reproductive-related vital rates (i.e., maturity

[Mat] and reproductive rate) with increasing age

truncation (see Appendix C). However, moving-window

correlations between growth rates and the elasticity

values showed that the overall contribution of changes

in the vital rates to changes in the growth rate was non-

stationary, displaying a different temporal pattern for

each subpopulation (Fig. 5b). For the north deme, the

correlation between rRI and its elasticity to changes in

Mat shifted from negative and significant in the first part

of the study period to positive and significant in the

second part, a similar pattern observed in the correlation

between r and its elasticity to the reproductive rate (Fig.

5b). This indicated that the switch in the control of the

population growth from survival to reproductive rate

was primary mediated by the change in the maturity

rates. By contrast, while rRI at the south deme displayed

a high dependency on changes of Mat for the entire

period (Fig. 5b), the correlation between r and the

FIG. 5. (a) Temporal variation of realized growth rate (r, solid black line), recruitment-independent growth rate (rRI, dottedblack line) and the global recruitment success contribution (r – rRI, dashed gray line) calculated for the north (left-hand column)and the south (right-hand column) deme. (b) The 12-year moving-window correlation coefficients between growth rates (r in solidline and rRI in dotted line) and their elasticity values to reproduction-related rates (see Appendix C). Circles represent significantcorrelation coefficients (P , 0.05 accounting for autocorrelation). (c) Absolute change (mean 6 SE) of elasticity of r toreproductive rate (population dynamics attribute) due to recruitment success (ecological variable, white) and length at 50%probability of being mature (phenotypic variable, gray) for the entire period, before 1995 and after 1996. Asterisks representstatistical significance.

* P , 0.05; ** P , 0.01; *** P , 0.005.


elasticity to the reproductive rate shifted from positive

and significant in the first part of the study to

nonsignificant in the more recent years (Fig. 5b); an

opposed pattern to that observed for the north deme.

Thus, changes in heritable maturation traits had high

influence on the population performance in the south

deme independently of the demographic state, while a

change in the recruitment success could trigger the

change in the correlation between r and the elasticity to

the reproductive rate.

Significant moving-window correlation coefficients

between r and its elasticity to the reproductive rate

(Fig. 5b) were negatively correlated to significant density-

dependent parameters (b) estimated from the non-

stationary stock–recruitment relationships for the two

demes (binomial correlation, cor¼�0.71, P� 0.001 for

the north deme; cor ¼ �0.72, P , 0.05 for the south

deme). This suggests that density dependence is signifi-

cantly regulating the population (at stock–recruitment

level) when the reproductive rate does not drive the

population growth. This trade-off between density

dependence and reproductive rate influence was observed

in the two demes suggesting that it occurs irrespective of

the fishing effect on the demographic structure.

Finally, Hairston et al.’s framework was used to

investigate the effect of recruitment success (ecological

variable) and maturation schedules (phenotypic vari-

able) on reproductive rate and the elasticity of r to

reproductive rate—two population dynamics attributes

affected by ecological and evolutionary processes. The

additive form for all models showed lower AIC values

compared to the interactive form (see Appendix D).

Models fitted for the reproductive rate were mainly

affected by the recruitment success variability (see

Appendix D), while those fitted for the elasticity of r

indicated a different pattern for each deme (Fig. 4c).

Both recruitment success and maturation schedules

showed a significant effect on the elasticity at the north

deme for both the whole and the truncated period. By

contrast, maturation schedules variability was the

unique significant effect on the elasticity at the south

deme attending to the entire period, while an effect of

the recruitment success was observed for the truncated

period (Fig. 4c). This suggests that the effect of fishing

on a low plastic phenotype and ecological processes may

additively affect population growth in the north deme

through the contribution (i.e., elasticity) of the repro-

ductive rate. By contrast, changes in the phenotype may

reduce the demographic effect of fishing in the south

deme due to the influence of phenotypic variation on the

recruitment success.

DISCUSSION

Here we have demonstrated that the level of adaptive

phenotypic plasticity may influence how maturation

schedules respond to size-selective removal of fish

biomass. Our study adds to the understanding of how

phenotypic plasticity influences population dynamics

and the ability of populations to respond to natural and

anthropogenic drivers (Hutchings et al. 2007, Reed et al.

2010, 2011). We argue that populations with less plastic

but more adaptive phenotypes may display fishing-

induced additive effects of the phenotype and the

demographic change on the population dynamics. This

is supported by the results obtained for the north deme,

in which the detrimental effects of fishing on both

maturation schedules and recruitment success negatively

influenced reproductive rate with direct consequences

for the population growth. By contrast, populations

with more plastic phenotypes such as the south deme

may dampen the detrimental effects of the demographic

erosion due to the influence of phenotypically plastic

changes in the maturation schedules on the recruitment

success (see discussion hereafter).

As a general rule, heterogeneous environments favor

the evolution of phenotypic plasticity (Reed et al. 2011).

Thus, optimal phenotypes in complex environments

differ from optimal phenotypes in homogenous envi-

ronments (Tuljapurkar et al. 2009). Although our results

on the probabilistic maturation reaction norms support

that phenotypic variation of exploited species result

from the combination of harvest selection and natural

selection (Olsen and Moland 2011), the relative contri-

bution of each driver depends on the degree of

environmental heterogeneity (Marshall et al. 2010).

Phenotypes adapted to more homogeneous environ-

ments should be more prone to fishing-induced adaptive

modifications through earlier maturation. This is likely

the scenario in the north deme, which is affected by the

large scale hydrography in the Bay of Biscay and

displays a large spawning area in which the selective

effect of fishing may favor early–maturing relative to

late-maturing genotypes (e.g., Barot et al. 2004, Olsen et

al. 2004). By contrast, the changes in the maturation

schedules of the south deme were likely the result of a

plastic response in somatic growth. Reduced intraspe-

cific competition triggered by the removal of biomass

due to fishing (i.e., density-dependent somatic growth;

e.g., Trippel 1995, Helser and Almeida 1997) may be one

of the plausible underlying mechanisms, while a

potential influence of environmental variability cannot

be disregarded (Domınguez-Petit et al. 2008, Devine and

Heino 2011). The shape of the maturation reaction norm

of this subunit suggests that growth variability main-

tains maturing probability independent of age. Since

spawners of different age can occupy space and time for

reproduction differently (Hutchings and Myers 1993),

life history traits in the south deme are likely adapted to

maintain age spawners diversity to colonize the high

spatial and temporal environmental heterogeneity that

maximizes offspring survival.

We acknowledge, however, that maturation reaction

norms can fail in identifying growth-independent effects

of environmental variation (e.g., Morita et al. 2009,

Uusi-Heikkila et al. 2011). It is worth noticing that the

two demes analyzed here were affected by a similar


increasing pattern of mean temperature during the

whole study period (Hidalgo et al. 2012), while the

maturation schedules progressed in the opposed direc-

tion. We recognize that the sharp difference in the

maturation reaction norms in the north deme could be

partially influenced by the growth-independent effects of

increasing temperature. However, we consider its effect

to be marginal compared to that induced by the size-

selective removal due to fishing. In fact, recent studies on

cod in the North Sea (a neighboring area to the Bay of

Biscay) concluded that fishing-induced selection toward

earlier maturation is the most parsimonious explanation

for the maturation trends observed once the temperature

effect is accounted for (Wright et al. 2011, Neuheimer

and Grønkjær 2012).

The relative contribution of human and natural

drivers to the phenotypic response is inherently linked

to the relative importance of maturation schedules and

recruitment success on the reproductive rate (Marshall

et al. 2010), and this is supported by our results. Thus, if

variation in maturation schedules is more relevant, then

the fecundity–age relationship has a strong effect on the

reproductive success increasing the maturing-age depen-

dence. Under this situation fishing-induced phenotypic

and demographic changes may have additive effects on

the population dynamics, which is supported by our

results for the north deme. By contrast, in populations

(or subunits) in which life histories are more variable

(allowing the population to cope with the fluctuating

environment), recruitment success may be more impor-

tant for the population dynamics. In this case, life

history variation is better adapted to maximize recruit-

ment success and can dampen demographic effects of

fishing, which remains consistent with our results for the

south deme. This could also partially explain the non-

stationary pattern of population regulation depicted by

the stock recruitment models.

Density-dependent regulation (i.e., overcompensation

or saturation) requires food competition or/and preda-

tion affecting the survival of early life and pre-recruit

stages (Rose et al. 2001). Cannibalism on young age

classes is a known regulatory mechanism in the north

deme due to the spatial coexistence of different cohorts

on the broad continental shelf of the Bay of Biscay

(Mahe et al. 2007). This regulation was clearly removed

by the demographic erosion (Hidalgo et al. 2012), yet a

synergistic effect of increasing temperature in the North

Atlantic in recent decades could contribute to the

survival of early life stages in the north deme (Goikoet-

xea and Irigoien 2013). In addition to this ecological

effect, the natural selection resulting from cannibalism

on small fish historically acted in opposite direction to

the long-term harvesting selection on large fish. Togeth-

er, this could promote stabilizing selection for matura-

tion in the north deme (Carlson et al. 2007).

Contemporary removal of adults cannibalizing on

young fish could disrupt this stabilizing pattern and

magnify the directional selection toward earlier matu-

ration. By contrast, the phenotypic change in somatic

growth in the south deme caused individuals to mature

at younger age that could influence the population

regulation mechanisms. An increased contribution of

younger spawners may favor to colonize a higher

diversity of spatial and temporal environmental oppor-

tunities for reproduction proposes (Hutchings and

Myers 1993). This would stabilize density dependence

and increase the recruitment success variability, which

represents a key vital rate contributing to the absolute

population growth rate in the south compared to the

north deme. In addition, it is worth noticing that

density-dependent regulation could be also favored due

to a potential contraction of the spawning areas at lower

densities, which could strengthen competition for

resources (Rose et al. 2001).

Fishing-induced demographic erosion decreases the

overall dependence of population growth rate upon the

adult survival while it increases the dependence on the

reproductive rate (Rouyer et al. 2011, Hidalgo et al.

2012, Durant et al. 2013). However, within this scenario,

inter-annual variation of reproductive rate dependence

does not necessary lead to a correlated variation in the

growth rate. Since annual variation in reproductive rate

results from the interaction of changes in the phenotype

(i.e., maturity schedules) with variation in the recruit-

ment success, the resulting interplay may range from

additive to buffering effects on the population growth

rate. Here, we illustrate that the combination of

demographic truncation (i.e., reduction of the adult

component of a population) and adaptive juvenescence

(i.e., evolutionary change to mature at younger age) of

the spawning stock of the north deme triggered an

additive effect on the increasing dependence of the inter-

annual variability of population growth upon reproduc-

tive rate. However, the plastic response of the matura-

tion schedules of the south deme dampened such

reproductive rate dependence. Recent studies based on

long-term simulations suggest that fishing induced

evolution may have a low influence on population

growth rate at longer time scales (Kuparinen and

Hutchings 2012). Our study focuses on a shorter time

scale but the results remain consistent with the study by

Kuparinen and Hutchings (2012). Our findings further

show how the interaction of evolutionary and ecological

processes through the reproductive rate may contribute

to reduce the variation in population growth and

stabilizing it in the long term.

It is worth emphasizing that our study shows two

contrasting scenarios within the same population. This

highlights the importance of embracing ecological

processes and phenotypic variation in spatially complex

populations, especially for those ecosystems shaped by

spatial fragmentation of the environment or those

receiving individuals from different geographical origin.

Heterogeneous populations, characterized by complex

population structure and life history diversity, produce

more temporal stability because of the independent but


complementary dynamics among conspecific subpopu-

lations, so called biocomplexity (Hilborn et al. 2003,

Schindler et al. 2010). Demes within complex popula-

tions such as hake in the Atlantic do not differ in the

neutral genetic variation. However, differences in the

adaptive phenotypic traits suggest that adaptive evolu-

tion may be faster than neutral evolution and can act at

a smaller spatial scale (Conover et al. 2006). Thus,

divergence of local adaptations on a finer scale can be

found even if no clear difference in genetic structure can

be detected by microsatellites (Conover et al. 2006,

Hutchings et al. 2007), which is supported by our

results.

Concluding remarks, study limitations, and future

challenges

Knowledge about mechanisms leading to spatial

variation in the degree of phenotypic plasticity within

populations is critical for understanding the interplay of

evolutionary and ecological dynamics (Reed et al. 2011).

The nascent field of evolutionary demography represents

an emerging approach to investigate the consequences of

phenotypic variation on population dynamics/growth

(e.g., Kinnison et al. 2008, Ezard et al. 2009),

communities (e.g., Post et al. 2008, Carlson et al.

2011), ecosystems (Harmon et al. 2009, Palkovacs et

al. 2009) and the potential reciprocal feedbacks between

ecology and evolution. Here, we demonstrate the

necessity of expanding this approach to the case of

harvested species, especially to those displaying complex

spatial structures. Our study also highlights the impor-

tance of the reproductive rate for shaping the evolu-

tionary demography of exploited fish populations,

because fishing can affect both its phenotypic (i.e.,

maturity rate) and its ecological (i.e., recruitment

success) component.

It is worth noticing some limitations of our study as

well as future challenges. First, we acknowledge that

applying a time-varying approach to the maturation

information would have improved our study (i.e.,

calculating maturation reaction norms by cohort). Due

to our data restrictions, we simulated somatic growth

scenarios for the two contrasting periods, which was

used to illustrate the different contributions of plastic

and adaptive changes for each deme. Second, though we

demonstrate that density-dependent regulation changes

with time, future studies will benefit from including

density dependence explicitly in matrix models in a non-

stationary manner. This implies several methodological

modifications, not only applying a different density-

dependent structure to each year but also investigating

whether a different density-dependent structure might be

applied at different age classes within each year. Third,

further research will need to investigate the temporal

scale at which evolutionary demography processes

operate. Kuparinen and Hutchings (2012) showed that

a long-term population growth tends to stabilize around

zero growth independently of fishing-induced evolution-

ary change. However, our results demonstrate that at a

short temporal scale (i.e., management scale), the

relative contribution of evolutionary and ecological

mechanisms that stabilize population growth are highly

diverse, even within two genetically indistinguishable

subpopulations of the same species.

To conclude, the mounting evidence that fishing may

cause contemporary evolution calls for the implementa-

tion of a Darwinian fisheries management (Jørgensen et

al. 2007). Our study moves forward on this view and

underscores the necessity to understand how fishing

affects the trait-mediated interactions with ecological

processes. This remains a keystone for the functioning of

harvested marine ecosystems and for preserving their

services to the society. Thus, our study calls for a more

comprehensive management approach, in which species

are not only numerically assessed, but where ecological,

evolutionary, and environmental processes are also

integratively accounted for.

ACKNOWLEDGMENTS

M. Hidalgo received support from Marie Curie Intra-European fellowship (FP7-PEOPLE-IEF-2008, EuropeanCommission; project No 236549, EVOLHAKE project) andNorMER platform. We are grateful to Anna Gardmark for hervaluable suggestions on the manuscript and Thomas H. G.Ezard for earlier methodological suggestions. M. Hidalgospecially thanks Enric Massutı and Beatriz Morales-Nin fortheir kindly support during the stay at their labs. We areindebted to two anonymous reviewers for their very useful andvaluable comments and suggestions.

LITERATURE CITED

Anderson, C. N., C. H. Hsieh, S. A. Sandin, R. Hewitt, A.Hollowed, J. Beddington, R. M. May, and G. Sugihara.2008. Why fishing magnifies fluctuations in fish abundance.Nature 452:835–839.

Barot, S., M. Heino, L. O’Brien, and U. Dieckmann. 2004.Reaction norms for age and size at maturation: study of thelong-term trend (1970–1998) for Georges Bank and Gulf ofMaine cod stocks. Ecological Applications 14:1257–1271.

Carlson, S. M., E. Edeline, L. A. Vøllestad, T. O. Haugen, I. J.Winfield, J. M. Fletcher, J. B. James, and N. C. Stenseth.2007. Four decades of opposing natural and human-inducedartificial selection acting on Windermere pike (Esox lucius).Ecology Letters 10:512–521.

Carlson, S. M., T. P. Quinn, and A. P. Hendry. 2011. Eco-evolutionary dynamics in Pacific salmon. Heredity 106:438–447.

Caswell, H. 2001. Matrix population models: construction,analysis, and interpretation. Sinauer Associates, Sunderland,Massachusetts, USA.

Conover, D. O., L. M. Clarke, S. B. Munch, and G. N. Wagner.2006. Spatial and temporal scales of adaptive divergence inmarine fishes and implications for conservation. Journal ofFish Biology 69:21–47.

Devine, J. A., and M. Heino. 2011. Investigating the drivers ofmaturation dynamics in Barents Sea haddock (Melanogram-mus aeglefinus). Fisheries Research 110:441–449.

Dieckmann, U., and M. Heino. 2007. Probabilistic maturationreaction norms: their history, strengths, and limitations.Marine Ecology Progress Series 335:253–269.

Domınguez-Petit, R., M. Korta, F. Saborido-Rey, H. Murua,M. Sainza, and C. Pineiro. 2008. Changes in size at maturityof European hake Atlantic populations in relation with stock


structure and environmental regimes. Journal of MarineSystems 71:260–278.

Durant, J. M., M. Hidalgo, T. Rouyer, D. Ø. Hjermann, L.Ciannelli, A. M. Eikeset, N. Yaragina, and N. C. Stenseth.2013. Population growth across heterogeneous environments:effects of harvesting and age structure. Marine EcologyProgress Series 480:277–287.

Ellner, S. P., M. A. Geber, and N. G. Hairston, Jr. 2011. Doesrapid evolution matter? Measuring the rate of contemporaryevolution and its impacts on ecological dynamics. EcologyLetters 14:603–614.

Enberg, K., C. Jørgensen, and M. Mangel. 2010. Fishing-induced evolution and changing reproductive ecology of fish:the evolution of steepness. Canadian Journal of Fisheries andAquatic Science 67:1708–1719.

Ezard, T. H. G., S. D. Cote, and F. Pelletier. 2009. Ecoevolu-tionary dynamics: disentangling phenotypic, environmentaland population fluctuations. Philosophical Transactions ofthe Royal Society B 364:1491–1498.

Ezard, T. H. G., J.-M. Gaillard, M. J. Crawley, and T.Coulson. 2008. Habitat dependence and correlations betweenelasticities of long-term growth rates. American Naturalist172:424–430.

Fisher, J. A. D., K. T. Frank, and W. C. Leggett. 2010. Globalvariation in marine fish body size and its role in biodiversity–ecosystem functioning. Marine Ecology Progress Series405:1–13.

Goikoetxea, N., and X. Irigoien. 2013. Links between therecruitment success of northern European hake (Merlucciusmerluccius L.) and a regime shift on the NE Atlanticcontinental shelf. Fisheries Oceanography 22:459–456.

Goodwin, N. B., A. Grant, A. L. Perry, N. K. Dulvy, and J. D.Reynolds. 2006. Life history correlates of density-dependentrecruitment in marine fishes. Canadian Journal of Fisheriesand Aquatic Science 63:494–509.

Hairston, N. G., Jr, S. P. Ellner, M. A. Geber, T. Yoshida, andJ. A. Fox. 2005. Rapid evolution and the convergence ofecological and evolutionary time. Ecology Letters 8:1114–1127.

Harmon, L. J., B. Matthews, S. DesRoches, J. Chase, J. Shurin,and D. Schluter. 2009. Evolutionary diversification instickleback affects ecosystem functioning. Nature 458:1167–1170.

Heino, M., U. Dieckmann, and O. R. Godø. 2002. Measuringprobabilistic reaction norms for age and size at maturity.Evolution 56:669–678.

Helser, T. E., and F. P. Almeida. 1997. Density-dependentgrowth and sexual maturity of silver hake in the north-westAtlantic. Journal of Fish Biology 51:607–623.

Hidalgo, M., et al. 2012. Context-dependent interplays betweentruncated demographies and climate variation shape thepopulation growth rate of a harvested species. Ecography35:637–649.

Hilborn, R., T. P. Quinn, D. E. Schindler, and D. E. Rogers.2003. Biocomplexity and fisheries sustainability. Proceedingsof the National Academy of Sciences USA 100:6564–6568.

Hsieh, C. H., A. Yamauchi, T. Nakazawa, and W. F. Wang.2010. Fishing effects on age and spatial structures underminepopulation stability of fishes. Aquatic Science 72:165–178.

Hutchings, J. A., and R. A. Myers. 1993. Effect of age on theseasonality of maturation and spawning of Atlantic cod,Gadus morhua, in the Northwest Atlantic. Canadian Journalof Fisheries and Aquatic Science 50:2468–2474.

Hutchings, J. A., D. P. Swain, S. Rowe, J. D. Eddington, V.Puvanendran, and J. A. Brown. 2007. Genetic variation inlife-history reaction norms in a marine fish. Proceedings ofthe Royal Society B 274:1693–1699.

ICES. 2009. Report of the Working Group on the Assessmentof Southern Shelf Stocks of Hake, Monk, and Megrim[WGHMM]. 5–11 May 2009. ICES CM 2009/ACOM:08.

International Council for the Exploration of the Sea,Copenhagen, Denmark.

Jørgensen, C., et al. 2007. Managing evolving fish stocks.Science 318:1247–1248.

Kareiva, P. 1990. Population dynamics in spatially complexenvironments: theory and data. Philosophical Transactionsof the Royal Society B 330:175–190.

Kinnison, M. T., M. J. Unwin, and T. P. Quinn. 2008. Eco-evolutionary vs habitat contributions to invasion in salmon:experimental evaluation in the wild. Molecular Ecology17:405–414.

Kuparinen. A., and J. A. Hutchings. 2012. Consequences offisheries-induced evolution for population productivity andrecovery potential. Proceedings of the Royal Society B279:2571–2579.

Law, R. 2000. Fishing, selection, and phenotypic evolution.ICES Journal of Marine Science 57:659–668.

Leslie, P. H. 1945. On the use of matrices in certain populationmathematics. Biometrika 33:183–222.

Lorenzen, K., and K. Enberg. 2002. Density-dependent growthas a key mechanism in the regulation of fish populations:evidence from among-population comparisons. Proceedingsof the Royal Society B 269:49–54.

Mahe, K., R. Amara, T. Bryckaert, M. Kacher, and J. M.Brylinski. 2007. Ontogenetic and spatial variation in the dietof hake (Merluccius merluccius) in the Bay of Biscay and theCeltic Sea. ICES Journal of Marine Science 64:1210–1219.

Marshall, C. T., and B. J. McAdam. 2007. Integratedperspectives on genetic and environmental effects onmaturation can reduce potential for errors of inference.Marine Ecology Progress Series 335:301–310.

Marshall, D. J., S. S. Hepell, S. B. Munch, and R. R. Warner.2010. The relationship between maternal phenotype andoffspring quality: Do older mothers really produce the bestoffspring? Ecology 91:2862–2873.

McClatchie, S., R. Goericke, G. Auad, and K. Hill. 2010. Re-assessment of the stock–recruit and temperature–recruitrelationships for Pacific sardine (Sardinops sagax). CanadianJournal of Fisheries and Aquatic Science 67:1782–1790.

Minto, C., R. A. Myers, and W. Blanchard. 2008. Survivalvariability and population density in fish populations. Nature452:344–348.

Morita, K., J. Tsuboi, and T. Nagasawa. 2009. Plasticity inprobabilistic reaction norms for maturation in a salmonidfish. Biology Letters 5:628–631.

Murua, H. 2010. The biology and fisheries of European hake,Merluccius merluccius, in the north-east Atlantic. Advancesin Marine Biology 58:97–154.

Neuheimer, A. B., and P. Grønkjær. 2012. Climate effects onsize-at-age: growth in warming waters compensates forearlier maturity in an exploited marine fish. Global ChangeBiology 18:1812–1822.

Ohlberger, J., Ø. Langangen, E. Edeline, E. M. Olsen, I. J.Winfield, J. B. James, J. Fletcher, N. C. Stenseth, and L. A.Vøllestad. 2011. Pathogen-induced rapid evolution in avertebrate life-history trait. Proceedings of the Royal SocietyB 278:35–41.

Oli, M. K., and B. Zinner. 2001. Partial life cycle analysis: amodel for birth-pulse populations. Ecology 82:1180–1190.

Olsen, E. M., M. Heino, G. R. Lilly, M. J. Morgan, J. Brattey,B. Ernande, and U. Dieckmann. 2004. Maturation trendsindicative of rapid evolution preceded the collapse ofnorthern cod. Nature 428:932–935.

Olsen, E. M., and E. Moland. 2011. Fitness landscape ofAtlantic cod shaped by harvest selection and naturalselection. Evolutionary Ecology 25:695–710.

Olsen, E. M., G. Ottersen, M. Llope, K.-S. Chan, G. Beaugran,and N. C. Stenseth. 2011. Spawning stock and recruitment innorth sea cod shaped by food and climate. Proceedings of theRoyal Society B 278:504–510.


Ottersen, G. 2008. Pronounced long-term juvenation in thespawning stock of Arcto-Norwegian cod and possibleconsequences for recruitment. Canadian Journal of Fisheriesand Aquatic Science 65:523–534.

Palkovacs, E. P., et al. 2009. Experimental evaluation ofevolution and coevolution as agents of ecosystem change inTrinidadian streams. Philosophical Transactions of theRoyal Society B 364:1617–1628.

Pelletier, F., T. Clutton-Brock, J. Pemberton, S. Tuljapurkar,and T. Coulson. 2007. The evolutionary demography ofecological change: linking trait variation and populationgrowth. Science 315:1571–1574.

Pita, A., M. Perez, S. Cervino, and P. Presa. 2011. What cangene flow and recruitment dynamics tell us about connectiv-ity between European hake stocks in the Eastern NorthAtlantic? Continental and Shelf Research 31:376–387.

Post, D. M., E. P. Palkovacs, E. G. Schielke, and S. I. Dodson.2008. Intraspecific phenotypic variation in a predator affectscommunity structure and cascading trophic interactions.Ecology 89:2019–2032.

Price, T. D., A. Qvarnstrom, and D. E. Irwin. 2003. The role ofphenotypic plasticity in driving genetic evolution. Proceed-ings of the Royal Society B 270:1433–1440.

Pyper, B., and B. Peterman. 1998. Comparison of methods toaccount for autocorrelation in correlation analyses of fishdata. Canadian Journal of Fisheries and Aquatic Science55:2127–2140.

R Development Core Team. 2010. R 2.12.1. R Project forStatistical Computing, Vienna, Austria. www.r-project.org

Reed, T. E., D. E. Schindler, and R. S. Waples. 2011.Interacting effects of phenotypic plasticity and evolution onpopulation persistence in a changing climate. ConservationBiology 25:56–63.

Reed, T. E., R. S. Waples, D. E. Schindler, J. J. Hard, andM. T. Kinnison. 2010. Phenotypic plasticity and populationviability: the importance of environmental predictability.Proceedings of the Royal Society B 277:3391–3400.

Rose, K. A., J. H. Cowan, K. O. Winemiller, R. M. Myers, andR. Hilborn. 2001. Compensatory density dependence in fishpopulations: importance, controversy, understanding andprognosis. Fish and Fisheries 2:293–327.

Rouyer, T., G. Ottersen, J. M. Durant, M. Hidalgo, D. Ø.Hjermann, J. Persson, L. C. Stige, and N. C. Stenseth. 2011.Shifting dynamic forces in fish stock fluctuations triggered byage truncation? Global Change Biology 17:3046–3057.

Sanchez, F., and J. Gil. 2000. Hydrographic mesoscalestructures and poleward current as a determinant of hake(Merluccius merluccius) recruitment in southern Bay ofBiscay. ICES Journal of Marine Science 57:152–170.

Schindler, D., R. Hilborn, B. Chasco, C. P. Boatright, T. P.Quinn, L. A. Rogers, and M. S. Webster. 2010. Populationdiversity and the portfolio effect in an exploited species.Nature 465:609–613.

Swain, D. P., A. F. Sinclair, and J. M. Hanson. 2007.Evolutionary response to size-selective mortality in anexploited fish population. Proceedings of the Royal SocietyB 274:1015–1022.

Trippel, E. A. 1995. Age at maturity as a stress indicator infisheries. BioScience 45:759–771.

Tuljapurkar, S., J. M. Gaillard, and T. Coulson. 2009. Fromstochastic environments to life histories and back. Philo-sophical Transactions of the Royal Society B 364:1499–1509.

Uusi-Heikkila, S., A. Kuparinen, C. Wolter, T. Meinelt, A. C.O’Toole, and R. Arlinghaus. 2011. Experimental assessmentof the probabilistic maturation reaction norm: conditionmatters. Proceedings of the Royal Society B 278:709–717.

Venturelli, P. A., C. A. Murphy, B. J. Shuter, T. A. Johnston,P. J. de Groot, P. T. Boag, J. M. Casselman, R.Montgomerie, M. D. Wiegand, and W. C. Leggett. 2010.Maternal influences on population dynamics: evidence froman exploited freshwater fish. Ecology 91:2003–2012.

Wright, P. J., C. P. Millar, and F. M. Gibb. 2011. Intrastockdifferences in maturation schedules of Atlantic cod Gadusmorhua. ICES Journal of Marine Science 68:1918–1927.

SUPPLEMENTAL MATERIAL

Appendix A

Sensitivity analysis of maturation reaction norms to a range of somatic growth rates scenarios (Ecological ArchivesA024-063-A1).

Appendix B

Description, a figure, and a table presenting the stock–recruitment model selection (Ecological Archives A024-063-A2).

Appendix C

A figure presenting the elasticity of growth rate to survival and reproductive-related vital rates (Ecological ArchivesA024-063-A3).

Appendix D

A table presenting the model selection for the contribution of ecological and evolutionary processes (Ecological ArchivesA024-063-A4).


http://www.esapubs.org/archive/appl/A024/063/







Integrated STEM & Career Education for Middle Schools

Documents